8b10b encoder and decoder working

Tested on all 24-bit inputs after sync and on ~500M of random input
with and without intermediate sync

7 files changed, 272 insertions, 66 deletions

diff --git a/fw/.gitignore b/fw/.gitignore
index 3e36fab..ed41bfa 100644
--- a/fw/.gitignore
+++ b/fw/.gitignore
@@ -8,4 +8,5 @@
 sources.c
 sources.tar.xz
 sources.tar.xz.zip
-8b10b_test
+8b10b_test_decode
+8b10b_test_encode
diff --git a/fw/8b10b.c b/fw/8b10b.c
index 550e87c..c5a52ca 100644
--- a/fw/8b10b.c
+++ b/fw/8b10b.c
@@ -1,39 +1,42 @@
 
+#include <assert.h>
+#include <stdio.h>
+
 #include "8b10b.h"
 
-static const struct entry_5b6b map_5b6b[32] = {
-    {0b100111, 0b011000}, /* D.00 */
-    {0b011101, 0b100010}, /* D.01 */
-    {0b101101, 0b010010}, /* D.02 */
-    {0b110001, 0b110001}, /* D.03 */
-    {0b110101, 0b001010}, /* D.04 */
-    {0b101001, 0b101001}, /* D.05 */
-    {0b011001, 0b011001}, /* D.06 */
-    {0b111000, 0b000111}, /* D.07 */
-    {0b111001, 0b000110}, /* D.08 */
-    {0b100101, 0b100101}, /* D.09 */
-    {0b010101, 0b010101}, /* D.10 */
-    {0b110100, 0b110100}, /* D.11 */
-    {0b001101, 0b001101}, /* D.12 */
-    {0b101100, 0b101100}, /* D.13 */
-    {0b011100, 0b011100}, /* D.14 */
-    {0b010111, 0b101000}, /* D.15 */
-    {0b011011, 0b100100}, /* D.16 */
-    {0b100011, 0b100011}, /* D.17 */
-    {0b010011, 0b010011}, /* D.18 */
-    {0b110010, 0b110010}, /* D.19 */
-    {0b001011, 0b001011}, /* D.20 */
-    {0b101010, 0b101010}, /* D.21 */
-    {0b011010, 0b011010}, /* D.22 */
-    {0b111010, 0b000101}, /* D.23 */
-    {0b110011, 0b001100}, /* D.24 */
-    {0b100110, 0b100110}, /* D.25 */
-    {0b010110, 0b010110}, /* D.26 */
-    {0b110110, 0b001001}, /* D.27 */
-    {0b001110, 0b001110}, /* D.28 */
-    {0b101110, 0b010001}, /* D.29 */
-    {0b011110, 0b100001}, /* D.30 */
-    {0b101011, 0b010100}  /* D.31 */
+static const struct entry map_5b6b[32] = {
+    {0b100111, 0b011000, 2}, /* D.00 */
+    {0b011101, 0b100010, 2}, /* D.01 */
+    {0b101101, 0b010010, 2}, /* D.02 */
+    {0b110001, 0b110001, 0}, /* D.03 */
+    {0b110101, 0b001010, 2}, /* D.04 */
+    {0b101001, 0b101001, 0}, /* D.05 */
+    {0b011001, 0b011001, 0}, /* D.06 */
+    {0b111000, 0b000111, 0}, /* D.07 */
+    {0b111001, 0b000110, 2}, /* D.08 */
+    {0b100101, 0b100101, 0}, /* D.09 */
+    {0b010101, 0b010101, 0}, /* D.10 */
+    {0b110100, 0b110100, 0}, /* D.11 */
+    {0b001101, 0b001101, 0}, /* D.12 */
+    {0b101100, 0b101100, 0}, /* D.13 */
+    {0b011100, 0b011100, 0}, /* D.14 */
+    {0b010111, 0b101000, 2}, /* D.15 */
+    {0b011011, 0b100100, 2}, /* D.16 */
+    {0b100011, 0b100011, 0}, /* D.17 */
+    {0b010011, 0b010011, 0}, /* D.18 */
+    {0b110010, 0b110010, 0}, /* D.19 */
+    {0b001011, 0b001011, 0}, /* D.20 */
+    {0b101010, 0b101010, 0}, /* D.21 */
+    {0b011010, 0b011010, 0}, /* D.22 */
+    {0b111010, 0b000101, 2}, /* D.23 */
+    {0b110011, 0b001100, 2}, /* D.24 */
+    {0b100110, 0b100110, 0}, /* D.25 */
+    {0b010110, 0b010110, 0}, /* D.26 */
+    {0b110110, 0b001001, 2}, /* D.27 */
+    {0b001110, 0b001110, 0}, /* D.28 */
+    {0b101110, 0b010001, 2}, /* D.29 */
+    {0b011110, 0b100001, 2}, /* D.30 */
+    {0b101011, 0b010100, 2}  /* D.31 */
 };
 
 static const int8_t map_6b5b[64] = {
@@ -87,22 +90,22 @@ static const int8_t map_6b5b[64] = {
     [0b111010] = 23  /* rd = -1 */
 };
 
-static const struct entry_5b6b K28 = {0b001111, 0b110000};
-
-static const struct entry_3b4b map_d_3b4b[7] = {
-    {0b1011, 0b0100}, /* D.x.0 */
-    {0b1001, 0b1001}, /* D.x.1 */
-    {0b0101, 0b0101}, /* D.x.2 */
-    {0b1100, 0b0011}, /* D.x.3 */
-    {0b1101, 0b0010}, /* D.x.4 */
-    {0b1010, 0b1010}, /* D.x.5 */
-    {0b0110, 0b0110}  /* D.x.6 */
+static const struct entry K28 = {0b001111, 0b110000};
+
+static const struct entry map_3b4b_d[8] = {
+    {0b1011, 0b0100, 2}, /* D.x.0  */
+    {0b1001, 0b1001, 0}, /* D.x.1  */
+    {0b0101, 0b0101, 0}, /* D.x.2  */
+    {0b1100, 0b0011, 0}, /* D.x.3  */
+    {0b1101, 0b0010, 2}, /* D.x.4  */
+    {0b1010, 0b1010, 0}, /* D.x.5  */
+    {0b0110, 0b0110, 0}, /* D.x.6  */
+    {0b1110, 0b0001, 2}  /* D.x.P7 */
 };
 
-static const struct entry_3b4b  Dx_P7 = {0b1110, 0b0001},
-                                Dx_A7 = {0b0111, 0b1000};
+static const struct entry Dx_A7 = {0b0111, 0b1000, 2};
 
-static const struct entry_3b4b map_3b4b_k[8] = {
+static const struct entry map_3b4b_k[8] = {
     {0b1011, 0b0100}, /* K.x.0 */
     {0b0110, 0b1001}, /* K.x.1 */
     {0b1010, 0b0101}, /* K.x.2 */
@@ -147,12 +150,12 @@ static const uint16_t k_sym_map[K_CODES_LAST] = {
     [K30_7] = 0b0111101000
 };
 
-void xfr_8b10b_reset(struct state_8b10b *st) {
+void xfr_8b10b_reset(struct state_8b10b_dec *st) {
     st->rx = 0;
     st->bit_ctr = 0; /* unsynchronized */
 }
 
-int xfr_8b10b_feed_bit(struct state_8b10b *st, int bit) {
+int xfr_8b10b_feed_bit(struct state_8b10b_dec *st, int bit) {
     uint32_t pattern = st->rx = (st->rx<<1 | !!bit) & 0x3ff;
     uint16_t comma = k_sym_map[K28_1];
     if (pattern == comma || pattern == ((~comma)&0x3ff)) {
@@ -184,7 +187,44 @@ int xfr_8b10b_feed_bit(struct state_8b10b *st, int bit) {
     return -DECODING_IN_PROGRESS;
 }
 
-bool xfr_8b10b_has_sync(struct state_8b10b *st) {
+bool xfr_8b10b_has_sync(struct state_8b10b_dec *st) {
     return st->bit_ctr != 0;
 }
 
+void xfr_8b10b_encode_reset(struct state_8b10b_enc *st) {
+    st->rd = -1;
+}
+
+int xfr_8b10b_encode(struct state_8b10b_enc *st, int data) {
+    if (data < 0) {
+        if (-data >= sizeof(k_sym_map)/sizeof(k_sym_map[0]))
+            return -EINVAL;
+
+        return k_sym_map[-data];
+    }
+
+    if (data > 255)
+        return -EINVAL;
+
+    int p5b = data&0x1f, p3b = data>>5;
+    //fprintf(stderr, "\nrd %d data %x p5b %d p3b %d\n", st->rd, data, p5b, p3b);
+
+    int x5b = (st->rd == -1) ?  map_5b6b[p5b].rd_neg : map_5b6b[p5b].rd_pos;
+    st->rd -= map_5b6b[p5b].disp * st->rd;
+    //fprintf(stderr, "\nnow: rd %d data %x p5b %d p3b %d\n", st->rd, data, p5b, p3b);
+    assert(st->rd == -1 || st->rd == 1);
+
+    int x3b = (st->rd == -1) ?  map_3b4b_d[p3b].rd_neg : map_3b4b_d[p3b].rd_pos;
+    if (p3b == 7) {
+        if (   (st->rd == -1 && (p5b == 17 || p5b == 18 || p5b == 20))
+            || (st->rd ==  1 && (p5b == 11 || p5b == 13 || p5b == 14))) {
+            //fprintf(stderr, "A7 override\n");
+            x3b = (st->rd == -1) ?  Dx_A7.rd_neg : Dx_A7.rd_pos;
+        }
+    }
+    st->rd -= map_3b4b_d[p3b].disp * st->rd; /* D.x.A7 and D.x.P7 both have parity 2 */
+    assert(st->rd == -1 || st->rd == 1);
+
+    return x5b<<4 | x3b;
+}
+
diff --git a/fw/8b10b.h b/fw/8b10b.h
index 84763b1..090faaf 100644
--- a/fw/8b10b.h
+++ b/fw/8b10b.h
@@ -3,6 +3,7 @@
 
 #include <stdint.h>
 #include <stdbool.h>
+#include <errno.h>
 
 enum k_code {
     K28_0=1, K28_1, K28_2, K28_3,
@@ -17,22 +18,24 @@ enum decoder_return_codes {
     DECODING_IN_PROGRESS
 };
 
-struct entry_3b4b {
-    uint8_t rd_neg, rd_pos;
+struct entry {
+    uint8_t rd_neg, rd_pos, disp;
 };
 
-struct entry_5b6b {
-    uint8_t rd_neg, rd_pos;
-};
-
-struct state_8b10b {
+struct state_8b10b_dec {
     uint32_t rx; 
     int bit_ctr;
 };
 
+struct state_8b10b_enc {
+    int rd;
+};
+
+void xfr_8b10b_reset(struct state_8b10b_dec *st);
+int xfr_8b10b_feed_bit(struct state_8b10b_dec *st, int bit);
+bool xfr_8b10b_has_sync(struct state_8b10b_dec *st);
 
-void xfr_8b10b_reset(struct state_8b10b *st);
-int xfr_8b10b_feed_bit(struct state_8b10b *st, int bit);
-bool xfr_8b10b_has_sync(struct state_8b10b *st);
+void xfr_8b10b_encode_reset(struct state_8b10b_enc *st);
+int xfr_8b10b_encode(struct state_8b10b_enc *st, int data);
 
 #endif
diff --git a/fw/8b10b_test.c b/fw/8b10b_test_decode.c
index d11e6b2..433f7b8 100644
--- a/fw/8b10b_test.c
+++ b/fw/8b10b_test_decode.c
@@ -21,8 +21,7 @@ static const char * const rc_names[] = {
 };
 
 int main(void) {
-    struct state_8b10b st;
-
+    struct state_8b10b_dec st;
     xfr_8b10b_reset(&st);
 
     int c;
@@ -33,14 +32,18 @@ int main(void) {
         if (comment) {
             if (c == '\n')
                 comment = 0;
+            printf("%c", c);
             continue;
         }
 
-        if (c == '\r' || c == ' ' || c == '\t' || c == '\n')
+        if (c == '\r' || c == ' ' || c == '\t' || c == '\n') {
+            printf("%c", c);
             continue;
+        }
 
         if (c == '#') {
             comment = 1;
+            printf("%c", c);
             continue;
         }
 
@@ -57,9 +60,13 @@ int main(void) {
         char sync_status = xfr_8b10b_has_sync(&st) ? 'S' : 'U';
 
         if (read_result >= 0) {
-            fprintf(stdout, "%c%02x ", sync_status, read_result);
+            //fprintf(stdout, "%c%02x ", sync_status, read_result);
+            fprintf(stdout, "%02x", read_result);
 
         } else {
+            if (-read_result == DECODING_IN_PROGRESS)
+                continue;
+
             if (-read_result > sizeof(rc_names)/sizeof(rc_names[0])) {
                 fprintf(stderr, "Illegal read result %d. Exiting.\n", read_result);
                 return 2;
@@ -71,7 +78,8 @@ int main(void) {
                 return 2;
             }
 
-            fprintf(stdout, "%c%s ", sync_status, msg);
+            //fprintf(stdout, "%c%s ", sync_status, msg);
+            fprintf(stdout, "%s", msg);
         }
     }
 }
diff --git a/fw/8b10b_test_encode.c b/fw/8b10b_test_encode.c
new file mode 100644
index 0000000..2ba75d9
--- /dev/null
+++ b/fw/8b10b_test_encode.c
@@ -0,0 +1,108 @@
+
+#include <stdio.h>
+#include <string.h>
+
+#include "8b10b.h"
+
+static const char * const rc_names[] = {
+    [K28_0] = "K.28.0",
+    [K28_1] = "K.28.1",
+    [K28_2] = "K.28.2",
+    [K28_3] = "K.28.3",
+    [K28_4] = "K.28.4",
+    [K28_5] = "K.28.5",
+    [K28_6] = "K.28.6",
+    [K28_7] = "K.28.7",
+    [K23_7] = "K.23.7",
+    [K27_7] = "K.27.7",
+    [K29_7] = "K.29.7",
+    [K30_7] = "K.30.7",
+};
+
+int hex_to_uint(const char *s, size_t len) {
+    if (len > 7)
+        return -2;
+
+    int acc = 0;
+    for (int i=0; i<len; i++) {
+        int digit = s[i];
+        if ('0' <= digit && digit <= '9')
+            digit -= '0';
+        else if ('a' <= digit && digit <= 'f')
+            digit -= 'a' - 0xa;
+        else if ('A' <= digit && digit <= 'F')
+            digit -= 'A' - 0xA;
+        else return -1;
+        acc = acc << 4 | digit;
+    }
+
+    return acc;
+}
+
+int main(void) {
+    struct state_8b10b_enc st;
+    xfr_8b10b_encode_reset(&st);
+
+    int c;
+    int comment = 0;
+    char parse_buf[32];
+    int parse_buf_pos = 0;
+    while ((c=fgetc(stdin)) != EOF) {
+        if (comment) {
+            if (c == '\n')
+                comment = 0;
+            printf("%c", c);
+            continue;
+        }
+
+        if (c == '#') {
+            comment = 1;
+            printf("%c", c);
+            continue;
+        }
+
+        if (c != '\r' && c != ' ' && c != '\t' && c != '\n') {
+            parse_buf[parse_buf_pos++] = (char)c;
+            if (parse_buf_pos == sizeof(parse_buf)) {
+                parse_buf[sizeof(parse_buf)-1] = '\0';
+                fprintf(stderr, "Parse error: Token \"%s[...]\" too long. Exiting.\n", parse_buf);
+                return 1;
+            }
+            continue;
+        }
+
+        if (parse_buf_pos == 0)
+            continue;
+
+        int data = 0;
+        for (int i=1; i<sizeof(rc_names)/sizeof(rc_names[0]); i++) {
+            if (!strncmp(rc_names[i], parse_buf, parse_buf_pos)) {
+                data = -i;
+                break;
+            }
+        }
+
+        if (!data) {
+            data = hex_to_uint(parse_buf, parse_buf_pos);
+            if (data < 0) {
+                parse_buf[parse_buf_pos] = '\0';
+                fprintf(stderr, "Invalid hex number: \"%s\"\n", parse_buf);
+                return 1;
+            }
+        }
+
+        parse_buf_pos = 0; /* reset for next token */
+
+        int encoded = xfr_8b10b_encode(&st, data);
+        if (encoded < 0) {
+            fprintf(stderr, "Invalid argument error encoding\n");
+            return 2;
+        }
+
+        for (int i=0; i<10; i++)
+            printf("%c", (encoded & (1<<(9-i))) ? '1' : '0');
+
+        printf("%c", c); /* print whitespace character */
+    }
+}
+
diff --git a/fw/8b10b_test_vector_gen.py b/fw/8b10b_test_vector_gen.py
new file mode 100755
index 0000000..12460c1
--- /dev/null
+++ b/fw/8b10b_test_vector_gen.py
@@ -0,0 +1,43 @@
+#!/usr/bin/env python3
+
+def parse_size(s):
+    s = s.lower()
+    SUFFIXES = {'k': 1e3, 'm': 1e6, 'g': 1e9}
+    if s[-1] in SUFFIXES:
+        return int(int(s[:-1]) * SUFFIXES[s[-1]])
+    return int(s)
+
+def hexdump(data, byte_per_line=64):
+    for i in range(0, len(data), byte_per_line):
+        out = data[i:i+byte_per_line]
+        print(' '.join(f'{out[k]:02x}' for k in range(len(out)))) # use len(out) to handle partial lines
+
+if __name__ == '__main__':
+    from itertools import product
+    import os
+    import argparse
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument('cmd')
+    parser.add_argument('-l', '--length', default='1M', help='Generate [length} byte test vector. Only applicable to random.')
+    parser.add_argument('-s', '--syncfreq', default=None, help='Emit comma every [syncfreq] bytes. Default: don\'t emit any commas. Only applicable to random.')
+    args = parser.parse_args()
+
+    length = parse_size(args.length)
+    syncfreq = parse_size(args.syncfreq)
+
+
+    if args.cmd == 'exhaustive_separated':
+        for i, j, k in product(range(256), range(256), range(256)):
+            print(f'K.28.1 {i:02x} {j:02x} {k:02x}')
+
+    elif args.cmd == 'exhaustive_block':
+        print('K.28.1')
+        for i, j, k in product(range(256), range(256), range(256)):
+            print(f'{i:02x} {j:02x} {k:02x}')
+
+    elif args.cmd == 'random':
+        for chunk in range(0, length, syncfreq):
+            print('K.28.1')
+            hexdump(os.urandom(min(length-chunk, syncfreq)))
+
diff --git a/fw/Makefile b/fw/Makefile
index 1098717..b779789 100644
--- a/fw/Makefile
+++ b/fw/Makefile
@@ -86,7 +86,10 @@ main.elf: main.o startup_stm32f030x6.o system_stm32f0xx.o $(HAL_PATH)/Src/stm32f
 program: main.elf openocd.cfg
 	openocd -f openocd.cfg -c "program $< verify reset exit"
 
-8b10b_test: 8b10b_test.c 8b10b.c
+8b10b_test_encode: 8b10b_test_encode.c 8b10b.c
+	gcc -o $@ $^
+
+8b10b_test_decode: 8b10b_test_decode.c 8b10b.c
 	gcc -o $@ $^
 
 clean: